Intelligent service recommendation processing method and device based on multi-device context awareness and server

Abstract

This invention discloses a method, apparatus, and server for intelligent service recommendation processing based on multi-device context awareness, relating to the fields of artificial intelligence and the Internet of Things. The method includes: collecting multi-source data; performing deep fusion and contextual understanding on the multi-source data to identify the real-time context of the target user; matching the identified real-time context of the target user with a pre-set scenario rule base; generating personalized service recommendation schemes and recommendations based on the current context when the current context reaches a threshold for triggering service recommendations; determining the optimal service presentation device; controlling the generation of personalized service recommendation schemes and recommendations, pushing strong reminder messages to the selected optimal service presentation device, and prominently displaying service suggestions and operation buttons. This invention has the advantages of enabling dynamic collaborative service recommendations across devices, real-time perception of complex user scenarios, providing seamless personalized services, and improving user experience and recommendation accuracy.

Description

Technical Field

[0001] This application relates to the fields of artificial intelligence and Internet of Things technology, and more specifically, to a method, apparatus, server, and computer-readable storage medium for intelligent service recommendation processing based on multi-device context awareness. Background Technology

[0002] In the current smart device service ecosystem, service recommendation mechanisms face multiple challenges. Service recommendation functions on devices like TVs and mobile phones have long been fragmented, failing to achieve dynamic cross-device collaboration. For example, when a user watches content on a TV, the system cannot trigger related service recommendations on the mobile device, such as dining or travel suggestions based on viewing behavior, leading to fragmented service scenarios. Recommendation logic relies excessively on static historical data or shallow keyword matching, lacking the ability to deeply analyze real-time scenarios. When users encounter complex scenarios such as flight delays and late-night arrivals, the system struggles to integrate multi-source information such as flight status, time, and location to identify potential late-night transportation needs, thus failing to provide accurate and timely services. The interaction process suffers from significant gaps; even when recommendations are generated, users still need to frequently switch to their mobile phones to complete transactions, severely weakening the interactive advantages of the large TV screen and compromising service continuity. More importantly, existing technologies lack forward-looking prediction mechanisms, unable to extrapolate subsequent chain reactions (such as transportation difficulties) from a single event (like flight delays) and proactively plan preventative solutions. These shortcomings mean that service recommendations remain at the level of passive response and cannot adapt to the dynamic and changing real-world needs of users. There is an urgent need to build an intelligent service framework that can deeply integrate multi-source data, perceive complex scenarios in real time, and achieve seamless collaboration between core home devices and personal terminals.

[0003] Therefore, existing technologies still need improvement and development. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method, device, server and storage medium for intelligent service recommendation processing based on multi-device context awareness, which addresses the problems and defects of the prior art. The present invention has the advantages of being able to realize dynamic collaborative service recommendation across devices, perceive user complex contexts in real time, provide seamless personalized services, and improve user experience and recommendation accuracy.

[0005] This application provides an intelligent service recommendation processing method based on multi-device context awareness, the technical solution of which is as follows: A method for intelligent service recommendation based on multi-device context awareness, comprising: Collect and acquire data from multiple sources, including: explicit user behavior data, device status data, environmental data, and data from third-party data sources; The multi-source data is deeply fused and contextualized to identify the real-time context in which the target user is located; The system identifies the target user's real-time context and matches it against a pre-defined context rule base to determine whether the current context reaches the threshold for triggering service recommendations. When the current scenario reaches the threshold for triggering service recommendations, a personalized service recommendation plan and recommendation text are generated based on the current scenario; and the optimal service presentation device is determined. The system will generate personalized service recommendations and descriptions, push strong reminder messages to selected devices that best present the service, and display service suggestions and operation buttons in a prominent manner.

[0006] The intelligent service recommendation processing method based on multi-device context awareness, wherein, after the steps of generating personalized service recommendation schemes and recommendations, pushing strong reminder messages to the selected optimal service presentation device, and displaying service suggestions and operation buttons in a prominent manner, the method further includes: When the system detects that the user has confirmed acceptance of the service operation instruction, it controls the invocation of the corresponding third-party service API to complete the transaction operation and synchronizes the transaction result to the preset user personal device; When the system receives a user's instruction to ignore an action, it records the instruction, feeds it back to the preset recommendation model to optimize subsequent recommendations, and then exits.

[0007] The aforementioned intelligent service recommendation processing method based on multi-device context awareness includes user explicit behavior data such as ticket purchase data and search data; environmental data such as time data and location data; and third-party data sources such as flight dynamic data and weather data.

[0008] The intelligent service recommendation processing method based on multi-device context awareness, wherein the step of deeply fusing and understanding the multi-source data to identify the real-time context of the target user includes: When the explicit user behavior data of the multi-source data is that the user booked a flight ticket through the first terminal, the control binds the flight information of the booked flight ticket to the first terminal; When the third-party data source of the multi-source data is the flight information obtained through a second terminal, and the flight is delayed beyond expectations; And when the device and environmental data from the multi-source data are used by the system to determine that the flight will arrive late at night based on time data and location data; The deep fusion and contextual understanding of the multi-source data are then performed to integrate the flight delay information and late-night arrival time data to identify the real-time scenario of the target user as arriving late at night and needing to be picked up from the airport.

[0009] The intelligent service recommendation processing method based on multi-device context awareness further includes the step of collecting and acquiring multi-source data as follows: In-depth analysis of family members’ historical service data across different devices, including booking records, browsing preferences, and payment habits, yields historical behavioral preference data corresponding to family members. The collection of implicit preference data for family members' devices includes the type of travel app, the level of ride-hailing they frequently use, and travel preference keywords mentioned by family members in social media or chat logs; Analyze the usage habits and behaviors of different family members on their devices, including the activity level and usage time of each user's terminal device; The historical service data, implicit preference data, and device usage habit data are stored as multidimensional preference data for family members.

[0010] The intelligent service recommendation processing method based on multi-device context awareness further includes the following steps: when the current context reaches the threshold for triggering service recommendation, a personalized service recommendation scheme and recommendation text are generated based on the current context; and the optimal service presentation device is determined. When the current scenario reaches the threshold for triggering service recommendations, a recommendation list containing a predetermined number of differentiated options is generated based on the multidimensional preference data, the current scenario, and the service types available from the third-party service API. Each option includes a brief service description, a price range, and a degree of matching with the preferences of a particular family member.

[0011] The intelligent service recommendation processing method based on multi-device context awareness, wherein the steps of controlling the generation of personalized service recommendation schemes and recommendations, pushing strong reminder messages to the selected optimal service presentation device, and displaying service suggestions and operation buttons in a prominent manner, further include: The control terminal's large screen pops up multiple service options, each displayed as an independent card, including the service name, key features, estimated price, and can selectively label recommended usernames and user preferences; Based on the degree of matching with user preferences, options that match the preferences of the majority or key decision-makers are intelligently recommended or highlighted; If no confirmation selection instruction is received within the predetermined time, the control will send a vote to assist in decision-making based on preset rules or through the terminal. When a confirmation of an option is detected, the system immediately calls the corresponding third-party service API to complete the transaction and synchronizes the transaction result to the mobile terminals of all relevant family members.

[0012] A smart service recommendation processing device based on multi-device context awareness, wherein the device includes: The multi-source data acquisition module is used to collect and acquire multi-source data, including: explicit user behavior data, device status data, environmental data, and data from third-party data sources; The context-aware module is used to perform deep fusion and context understanding of the multi-source data to identify the real-time context in which the target user is located. The scenario matching and matching judgment module is used to match the real-time scenario of the identified target user with the pre-set scenario rule library to determine whether the current scenario reaches the threshold for triggering service recommendations. The service decision module is used to generate personalized service recommendation schemes and recommendations based on the current scenario when the current scenario reaches the threshold for triggering service recommendations; and to determine the optimal service presentation device. The push control module is used to control the generation of personalized service recommendation schemes and recommendations, push strong reminder messages to selected optimal service presentation devices, and display service suggestions and operation buttons in a prominent manner.

[0013] A server includes a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by one or more processors, the one or more programs comprising the method for performing any one of the methods.

[0014] A computer-readable storage medium, wherein, when instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to perform any of the methods described above.

[0015] The beneficial effects of this invention are as follows: This invention provides a method, apparatus, server, and storage medium for intelligent service recommendation processing based on multi-device context awareness. By collecting multi-source data, deeply fusing and understanding contexts, matching rule bases, generating recommendations, and pushing them to the optimal device, this invention solves the problems of service fragmentation and interaction disconnect. It has the advantages of enabling dynamic collaborative service recommendations across devices, real-time perception of complex user scenarios, and providing seamless personalized services, thereby improving user experience and recommendation accuracy. Furthermore, this invention can achieve the following technical effects: 1) From "people looking for services" to "services looking for people": The system of this invention has foresight and can proactively provide solutions before users are even aware of their needs, thus achieving true intelligence.

[0016] 2) It can achieve deep integration of context: the recommendation is not based on isolated data, but on a multi-dimensional integrated understanding of behavior, environment and real-time events, resulting in extremely high recommendation accuracy.

[0017] 3) Enables a closed-loop transaction process on large screens: Fully leverages the display advantages of large TV screens and the decision-making atmosphere of the living room environment to allow discovery, decision-making, and payment to be completed on a single device, providing a smooth experience.

[0018] 4) Enables multi-device collaborative intelligence: The system of this invention cleverly allocates roles between mobile phones (collecting data and receiving notifications) and televisions (presenting and making decisions), forming a highly efficient collaboration of "mobile phone perception and television decision-making". Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the intelligent service recommendation processing method based on multi-device context awareness provided in Embodiment 1 of the present invention.

[0021] Figure 2 This is a schematic diagram illustrating the specific implementation process of the intelligent service recommendation processing method based on multi-device context awareness provided in Embodiment 2 of the present invention.

[0022] Figure 3 The present invention provides a schematic diagram of an embodiment of an intelligent service recommendation processing device based on multi-device context awareness.

[0023] Figure 4 This is a block diagram illustrating the internal structure of the server provided in an embodiment of the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0025] It should be noted that if the embodiments of the present invention involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of the components in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0026] Traditional smart device service recommendation systems suffer from significant deficiencies in multi-source data fusion mechanisms. Specifically, these deficiencies manifest in a lack of deep correlation processing between user behavior data, device status data, environmental data, and third-party data sources, leading to insufficient real-time scenario recognition accuracy. During scenario matching, the dynamic adaptation capability between the scenario rule base and the user's current state is weak, failing to effectively trigger service recommendation thresholds. At the device collaboration level, the decision-making logic for service presentation on large TV screens and mobile terminals is lacking, preventing the recommendation scheme from allocating optimal devices based on user interaction habits. Furthermore, the cross-device continuity of service transaction processes is not guaranteed, and the push mechanism for strong reminder messages and the highlighting strategy of the operation interface are not under unified control, thus affecting the system's performance on key performance indicators, including the timeliness of service recommendations, the completeness of user interaction paths, and the accuracy of scenario prediction.

[0027] For example, when a user queries flight information and completes a ticket booking through a smart TV terminal, the system collects the ticket purchase record from the user's explicit behavioral data and binds the flight information to the TV device. Real-time flight dynamic data transmitted from a third-party data source shows that the flight is delayed beyond expectations, and the time information in the environmental data further indicates that the arrival time will be in the late night. Location data confirms that the user's home address is located in the area surrounding the airport. However, in the existing technology, because the multi-source data is not deeply integrated, the system fails to contextually correlate the flight delay information with the late-night arrival time data, resulting in the failure of real-time scenario recognition and the inability to determine that the user is arriving late at night and needs to be picked up. At this time, the service recommendation threshold is not triggered, the TV screen does not generate a recommendation scheme that includes airport pickup services, nor does it decide on the optimal display device. The user is forced to interrupt the TV operation and manually search for airport pickup options using a mobile application, causing a break in the interaction process.

[0028] If the above problems are not resolved, it will be impossible to realize the proactive service recommendation function based on real-time scenarios. The service acquisition path for users in multi-device environments will be forced to be extended, and cross-terminal jumps in the transaction process will lead to fragmented interactive experiences. The lack of context awareness means that service recommendations rely solely on historical behavioral data and cannot respond to immediate needs caused by sudden events, thereby reducing the relevance and practicality of service recommendations. The failure of device collaboration mechanisms will prevent the service transaction potential of large TV screens from being released, and system resources will be redundantly consumed in the data collection and processing stages, ultimately affecting users' overall trust in and stickiness to the smart device service ecosystem.

[0029] In traditional technologies, the collection of multi-source data involves isolated storage without establishing semantic relationships between explicit user behavior data, device status data, environmental data, and data from third-party sources. The contextual recognition stage lacks the ability to understand the data's context, leading to discrepancies between real-time scenario judgments and actual user needs. The matching logic of the scenario rule base cannot dynamically assess whether the current scenario meets the service recommendation triggering conditions, and the threshold judgment mechanism is rigid. The generation of service recommendation schemes lacks personalized customization based on scenario characteristics, and the device's decision-making regarding recommendation text and interface presentation lacks a basis. The push control of strong reminder messages does not achieve multi-terminal collaboration, and the highlighting strategy for service suggestions and operation buttons fails to adapt to optimal device characteristics.

[0030] In response, this application proposes an intelligent service recommendation processing method based on multi-device context awareness.

[0031] like Figure 1 As shown, an intelligent service recommendation processing method based on multi-device context awareness according to Embodiment 1 of the present invention includes the following steps: Step S100: Collect and acquire multi-source data, including: user explicit behavior data, device status data, environmental data, and data from third-party data sources; Step S200: Perform deep fusion and contextual understanding on the multi-source data to identify the real-time scenario in which the target user is located; Step S300: Match the real-time scenario of the identified target user with the preset scenario rule base to determine whether the current scenario reaches the threshold for triggering service recommendation; Step S400: When the current scenario reaches the threshold for triggering service recommendation, a personalized service recommendation scheme and recommendation text are generated based on the current scenario; and the optimal service presentation device is determined. Step S500: The system will generate personalized service recommendation schemes and recommendations, push strong reminder messages to the selected optimal service presentation device, and display service suggestions and operation buttons in a prominent manner.

[0032] For ease of understanding, the following explains some key terms in this embodiment: The multi-device context awareness in this embodiment refers to the ability of the present invention to comprehensively analyze data from different devices (such as smartphones, smart TVs, smart speakers, etc.) and combine them with environmental information to fully and accurately understand the specific context in which the user is located.

[0033] The intelligent service recommendation refers to the system of the present invention proactively and intelligently recommending services that meet the user's needs based on the user's real-time context and preferences.

[0034] The multi-source data in this embodiment refers to various types of data collected by the system during context awareness and recommendation. These data come from diverse sources, including user behavior, device status, environmental information, and external third-party data.

[0035] In this embodiment, explicit user behavior data refers to behavioral data that users actively generate through their devices and can be directly observed, such as ticket purchase records, search queries, and application usage history.

[0036] The device's own status data refers to relevant information about the device's current operation or configuration, such as the device's geographical location, connection status, battery level, and screen display content.

[0037] The environmental data in this embodiment refers to relevant information about the user's physical or temporal environment, such as current time, date, weather, temperature, light intensity, noise level, etc.

[0038] The third-party data source refers to data provided by external service providers or platforms, such as flight status, traffic conditions, news information, social media information, etc.

[0039] The deep fusion and contextual understanding in this embodiment refer to the system's comprehensive analysis of the collected multi-source data. It is not just a simple data overlay, but rather an algorithm model that mines the correlation between data, understands the deeper meaning and user intent behind the data, and thus constructs a complete user scenario profile.

[0040] The real-time scenario in this embodiment refers to the specific state and needs of the user at a specific time and place, such as the user being traveling, resting at home, or preparing to travel.

[0041] The scenario rule base refers to a predefined set of rules used to describe the conditions and logic for triggering service recommendations under different scenarios. These rules can be built based on expert experience or machine learning models.

[0042] The threshold for triggering service recommendations in this embodiment refers to a quantitative standard or logical condition set in the scenario rule base to determine whether the current scenario meets the recommendation criteria. When the scenario meets this threshold, the system will initiate the service recommendation process.

[0043] The personalized service recommendation scheme and recommendation text in this embodiment refer to the service suggestions tailored by the system based on the user's real-time situation and personal preferences, accompanied by attractive explanatory text.

[0044] The optimal service presentation device refers to the device that the system intelligently selects based on the current context, user preferences, and device characteristics to display recommended services to the user, such as smart TVs, smartphones, and smart speakers.

[0045] The strong reminder message in this embodiment refers to a notification sent to the user in a high-priority and prominent manner (such as pop-up window, vibration, sound, etc.), which aims to ensure that the user can notice the recommended service in a timely manner.

[0046] The prominent display of service suggestions and operation buttons in this embodiment refers to showcasing service recommendations in a prominent, user-friendly, and easy-to-understand manner on the selected presentation device, while providing clear interactive options.

[0047] This invention provides an intelligent service recommendation processing mechanism based on multi-device context awareness. In specific implementation, the system is first configured to collect multi-source data. This multi-source data collection process is the foundation of intelligent service recommendation, and its purpose is to comprehensively acquire information related to the user's context. For example, data can be collected from the user's smartphone, smart TV, smart speaker, and other devices. Specifically, explicit user behavior data can be collected, such as clicks, browsing, and purchase records in a particular application; device status data can be collected, such as the device's geographical location information, currently running applications, and screen brightness; environmental data can be collected, such as indoor temperature and light intensity obtained through sensors, or external weather information obtained through the network; and data from third-party data sources can also be collected, such as flight delay information and traffic conditions obtained through public APIs (Application Programming Interfaces). This data can be collected in real-time or periodically and stored in a backend server or cloud platform.

[0048] Secondly, the system performs deep fusion and contextual understanding of the multi-source data to identify the real-time scenario of the target user. After data collection, the system needs to integrate and analyze this heterogeneous data. For example, it aligns data from different sources with timestamps and semantic associations, and uses machine learning algorithms (such as neural networks and decision trees) to perform pattern recognition and semantic analysis on the fused data. In this way, the system can extract meaningful information from fragmented data. For example, when it detects that a user's mobile phone calendar event shows "business trip," and the device location data indicates that the user is at the airport, and a third-party data source shows that the flight is about to take off, the system can understand that the user is in a real-time scenario of "about to travel for business."

[0049] Next, the system identifies the target user's real-time context and matches it against a pre-defined scenario rule base to determine if the current context meets the threshold for triggering service recommendations. Once a real-time context is identified, the system compares it with pre-defined rules. For example, the scenario rule base might contain rules such as "when a user is at an airport and their flight is delayed for more than 2 hours, a food and beverage service recommendation will be triggered." The system evaluates whether the current context meets the conditions defined by these rules. This matching process can be based on simple conditional judgments or involve complex logical reasoning. When the context meets the conditions set in the rules, it is considered to have reached the threshold for triggering service recommendations.

[0050] Furthermore, when the current scenario reaches the threshold for triggering service recommendations, a personalized service recommendation scheme and recommendation text are generated based on the current scenario; and the optimal service presentation device is determined. Once it is determined that service recommendations are needed, the system of this invention will generate customized service suggestions based on the currently identified real-time scenario, combined with information such as the user's historical preferences and behavioral habits. For example, if the scenario is "flight delay," this embodiment of the invention will recommend restaurants, lounge services, or delay insurance near the airport. At the same time, the system of this invention will also generate attractive recommendation texts for these recommended services. In addition, the system will intelligently select the device most suitable for displaying these recommended services. For example, if the user is watching a smart TV, a smart TV may be selected as the presentation device; if the user is using a mobile phone, the user's mobile phone may be selected. This decision can be based on factors such as device activity, screen size, and the user's current interaction focus.

[0051] Finally, the system generates personalized service recommendation schemes and descriptions, pushes strong reminder messages to the selected optimal service presentation device, and prominently displays service suggestions and operation buttons. After generating the recommendation scheme and determining the presentation device, the system sends a notification to the user through that device. For example, a full-screen recommendation window may pop up on a smart TV screen, or a notification with vibration or sound alerts may be displayed on a mobile phone. The recommended content is presented in a clear and easy-to-read manner, such as using large fonts, eye-catching colors, or a card-style layout. Simultaneously, the recommendation interface provides clear operation buttons, such as "Accept," "Ignore," and "View Details," facilitating quick user interaction.

[0052] The technical solution of the present invention will be described in more detail below through a more specific example: For example, User A plans to travel from location X to location Y by flight. Before departure, User A books the ticket using their smartphone. The system of this invention continuously collects multi-source data. Specifically, the system collects User A's ticket purchase data (explicit user behavior data) from User A's smartphone, which includes information such as flight number and departure time. Simultaneously, the system also continuously acquires the geographic location information of User A's smartphone (device status data), as well as real-time weather data and flight status data obtained through the network (third-party data sources).

[0053] While user A is en route to the airport, the system of this invention performs deep fusion and contextual understanding of this multi-source data to identify the real-time scenario in which the target user is located. For example, the system detects that user A's mobile phone location has entered the airport area, and obtains from a third-party flight dynamic data source that user A's flight is expected to be delayed by 3 hours due to weather conditions. Combining this information, the system of this embodiment of the invention understands that user A is in a real-time scenario of "waiting at the airport and the flight is severely delayed."

[0054] Subsequently, the system of this embodiment of the invention matches the identified real-time scenario of "waiting at the airport with a severely delayed flight" with a pre-set scenario rule base to determine whether the current scenario reaches the threshold for triggering service recommendations. The scenario rule base pre-sets a rule: "When a user is at an airport and their flight is delayed for more than 2 hours, a recommendation for dining or rest services will be triggered." Since user A's flight delay has reached 3 hours, this scenario meets the threshold set by the rule.

[0055] When the current situation reaches the threshold for triggering service recommendations, the system of this invention generates personalized service recommendation schemes and recommendations based on the current situation; and determines the optimal service presentation device. Based on user A's real-time situation (flight delay) and historical preferences (e.g., user A prefers coffee and light meals), the system generates service recommendation schemes such as "discounted meal at a coffee shop in the airport" and "airport VIP lounge experience voucher," accompanied by recommendations such as "Flight delayed, want to relax with a coffee?" or "Long wait, enjoy a comfortable rest." Simultaneously, the system determines that user A is currently browsing news on a smartphone, and therefore decides to use the smartphone as the optimal service presentation device.

[0056] Finally, the system control of this invention generates personalized service recommendations and messages, pushes strong reminder messages to the selected optimal service presentation devices, and prominently displays service suggestions and operation buttons. A full-screen notification immediately pops up on User A's smartphone screen, displaying "Flight delayed, we recommend: [Café Name] special offer, click to claim" and "VIP lounge experience voucher, click to view." This notification is presented prominently and includes operation buttons such as "Claim Now" and "View Later," guiding the user to the next step. Through this series of steps, this method proactively and intelligently provides personalized service recommendations at the moment when User A needs help most.

[0057] Based on the above examples, this method, through its overall technical concept, effectively solves the problems of service isolation, passive recommendation, broken interaction, and lack of foresight in existing smart device service recommendations, demonstrating a significant technical contribution.

[0058] Compared to the isolated nature of service recommendations in existing technologies, the embodiments of the present invention, by collecting and acquiring multi-source data, can integrate heterogeneous information from explicit user behavior, device status, environment, and third-party data sources. For example, in the aforementioned flight delay example, the system not only focuses on the user's ticket purchase records on their mobile phone but also combines various data such as device location and real-time flight status, breaking the limitations of a single device or a single data source and achieving cross-device and cross-data source information collaboration.

[0059] Regarding the passivity of recommendations, the method of this invention deeply fuses and understands multi-source data to identify the real-time context of the target user and matches it with a pre-set scenario rule base to determine whether the current context reaches the threshold for triggering service recommendations. This enables the system to proactively perceive user needs, rather than passively responding based solely on historical behavior or simple keywords. In the example, the system can anticipate the inconvenience that flight delays may cause and proactively provide solutions before the user generates a need, demonstrating its foresight and proactivity.

[0060] Furthermore, this method significantly improves the user experience in generating personalized service recommendations and descriptions, determining the optimal service presentation device, controlling the generation of personalized service recommendations and descriptions, pushing strong reminder messages to the selected optimal service presentation device, and displaying service suggestions and operation buttons in a prominent manner. By intelligently determining the optimal presentation device and displaying the information in a strong reminder and prominent manner, the timeliness and effectiveness of the recommendation information are ensured. In the example, the system directly pushes the recommendation information to the user's currently used smartphone and presents it in a prominent manner, avoiding the cumbersome process of switching between different devices, optimizing the interaction flow, and solving the problem of interaction fragmentation in existing technologies.

[0061] In summary, the embodiments of the present invention, through the overall design of multi-source data collection, deep context perception, intelligent matching and decision-making, and proactive push and optimized interaction, achieve accurate understanding and prediction of the user's real-time context, thereby providing more timely, personalized and seamless service recommendations, greatly improving the intelligence level of intelligent service recommendations and user satisfaction.

[0062] In some of the embodiments described above in this application, a method for generating and pushing intelligent service recommendation schemes is proposed. However, in the implementation process, only the service recommendation information is pushed to the user, and there is a lack of an effective response mechanism for the user's subsequent operations. This may result in the recommended services not being adopted in a timely manner or the user feedback not being effectively utilized by the system, thereby affecting the user experience and the continuous optimization of the recommendation system.

[0063] In response, this application further proposes that after the steps of generating personalized service recommendation schemes and recommendations, pushing strong reminder messages to selected optimal service presentation devices, and displaying service suggestions and operation buttons in a prominent form, the control also includes: when a user confirms acceptance of the service operation instruction, the control calls the corresponding third-party service API to complete the transaction operation and synchronizes the transaction result to the preset user's personal device; when a user selects to ignore the operation instruction, the control system records and feeds it back to the preset recommendation model to optimize subsequent recommendations, and then exits.

[0064] Specifically, detecting a user's confirmation of service acceptance means that after receiving a service recommendation, the user interacts with the service presentation device to explicitly express their willingness to accept the recommended service. The detection methods may include, but are not limited to: the user clicking on an operation button such as "Accept," "Confirm," or "Book Now" displayed on the service presentation device; the user explicitly expressing their willingness to accept via voice command, which is then recognized by the voice recognition module; or the user confirming via gesture recognition or eye tracking, which is then analyzed by the corresponding sensors and processing modules.

[0065] The control system invokes the corresponding third-party service API to complete the transaction operation. The third-party service API refers to an interface opened by an external system providing specific services (such as a ticketing platform, ride-hailing platform, hotel booking platform, etc.), allowing this system to interact with it programmatically and make function calls to realize the actual transaction or booking process of the recommended service. Its implementation may include, but is not limited to: The system in this embodiment selects the corresponding third-party service API based on the service type and provider information included in the recommendation scheme; sends a request containing transaction parameters (such as service ID, quantity, user payment information, etc.) to the third-party service API interface via HTTP / HTTPS protocol; receives and processes the transaction result returned by the third-party service API; and synchronizes the transaction result to a preset user personal device. The transaction result refers to the transaction status, order details, confirmation information, etc., returned by the third-party service system after the transaction operation is completed through the third-party service API. The user personal device refers to a mobile terminal device used by the user daily and associated with this system, such as a smartphone or tablet. Synchronization methods may include, but are not limited to: pushing transaction results to users' personal devices in real time via message push services (such as MQTT, WebSocket); sending transaction details to users via email or SMS services; storing transaction results in the cloud and allowing users' personal devices to access and query them through applications.

[0066] The "receiving user's instruction to ignore" refers to the user, after receiving a service recommendation, interacting with the service presentation device to explicitly express their intention not to accept or temporarily disregard the recommended service. The methods of receiving this instruction may include, but are not limited to: the user clicking on operation buttons such as "Ignore," "Cancel," "Later," or "Not Interested" displayed on the service presentation device; the user explicitly expressing their intention to ignore the service via voice command; or the user performing the ignore operation through specific gestures or eye-tracking commands.

[0067] The control system records feedback to a preset recommendation model to optimize subsequent recommendations. This recording refers to user feedback on the recommendation service, including whether the user accepted or ignored the request, the reason for ignoring (if provided by the user), and detailed contextual information for the current recommendation. The preset recommendation model is an algorithmic model used to generate personalized service recommendation schemes, such as collaborative filtering or deep learning recommendation models. Feedback methods may include, but are not limited to: using user feedback data as training samples to update the parameters of the recommendation model online or offline; using user feedback as feature input to adjust the weights or preference factors of the recommendation algorithm; and using reinforcement learning mechanisms to reward or punish recommendation strategies based on user behavior. Exit means that after the user chooses to ignore the recommendation service, the system ends the current service recommendation process and no longer displays or processes the recommendation. This can be implemented in ways including, but are not limited to: closing the recommendation pop-up or prompt information on the service presentation device; clearing temporary data related to the current recommendation; and restoring the system state to its state before the recommendation, awaiting the next contextual trigger.

[0068] This application's solution, after the intelligent service recommendation scheme and recommendation text are generated and pushed to the optimal service presentation device, further constructs a closed-loop processing mechanism for user feedback. When a user issues a confirmation instruction to accept the service through the service presentation device, the system of this invention can respond immediately, controlling the invocation of the corresponding third-party service API to complete the actual transaction operation, thereby transforming the virtual recommendation into actual service acquisition. After the transaction is completed, the system will promptly synchronize the transaction result to the preset user's personal device, ensuring that the user can obtain service confirmation information and details on their commonly used device, improving service convenience and user trust. On the other hand, if the user chooses to ignore the recommendation, the system receives the user's ignoring instruction and does not simply terminate the process, but instead feeds back the user feedback along with relevant records such as the recommendation context to the preset recommendation model. This feedback mechanism allows the recommendation model to learn user preferences and rejection patterns, thereby optimizing and adjusting subsequent recommendations, avoiding repeated recommendations of services that users are not interested in, and improving the accuracy of recommendations and user satisfaction. Through this two-way feedback processing, the solution of this application not only completes the closed loop of service recommendation but also achieves continuous adaptive optimization of the recommendation system, making the entire intelligent service recommendation system more intelligent, efficient, and user-friendly.

[0069] The following is a concrete example. Assume a user books a flight through a first terminal. The system, through deep fusion and contextual understanding, identifies that the user will arrive at their destination late at night and determines that the current scenario meets the threshold for triggering a service recommendation. Specifically, in this invention, the system determines that this scenario matches the service scenario of "arriving late at night requiring airport pickup" and meets the service trigger threshold, generating a recommended airport pickup service plan and pushing it to the user's smart TV as the optimal service presentation device. When the airport pickup service suggestion and operation button are prominently displayed on the smart TV screen, if the user clicks the "Book Now" button, the operation instruction is detected. At this time, the system will control the invocation of a preset third-party ride-hailing service API, such as Didi Chuxing or Deda Taxi, and input parameters such as the user's flight information, destination, and estimated arrival time to complete the airport pickup service booking transaction. After the transaction is successful, the system will synchronize the transaction results, such as the order number, driver information, and estimated fare, to the user's smartphone or other preset personal devices via push notification service. Conversely, if the user clicks the "Ignore" button on the screen, the ignore operation instruction is received. The system will then record detailed information about this recommendation (such as the recommended service type, recommendation text, user context, and user ignore behavior) and feed it back to the recommendation model in the background. This recommendation model can be a deep learning-based recommendation network. By using this feedback data as negative samples for training, the model parameters are adjusted so that in similar future scenarios, the system can more accurately determine whether the user needs airport pickup service or recommend other services that better match the user's preferences. After that, the recommendation process ends.

[0070] Through the above technical solution, this application can effectively capture and respond to user operation commands after the intelligent service recommendation scheme is pushed. When the user accepts the recommendation, the system can quickly call the third-party service API to complete the transaction, transforming the recommendation into actual service acquisition, greatly improving the immediacy and convenience of the service. At the same time, the synchronization of the transaction result ensures that the user can obtain service confirmation information in a timely manner, enhancing the user's trust in the system. When the user chooses to ignore the recommendation, the system does not simply terminate, but uses the user feedback data to optimize the recommendation model, enabling the recommendation system to continuously learn and improve, thereby improving the accuracy of subsequent recommendations and user satisfaction, avoiding invalid recommendations, and improving the intelligence level and user experience of the entire intelligent service recommendation system.

[0071] The present invention will be further described in detail below through another specific application embodiment: like Figure 2 As shown in the illustration, this specific application embodiment provides an intelligent service recommendation processing method based on multi-device context awareness, which includes the following steps: S10. Multi-source data input, including: Explicit user behaviors, such as family members booking airline tickets through television, where flight information is then linked to the family television; or mobile phone searches; Equipment and environmental data, including time data, location data, and equipment status data. This invention's system uses time, location, and other data to determine that a user will arrive late at night.

[0072] Third-party data sources include flight status and weather, such as the system learning of a serious flight delay via mobile phone.

[0073] Then proceed to S20; S20. Perform situational perception and judgment, including: Data fusion and contextual understanding are performed, such as using a context-aware engine to fuse data such as flight delay information and late-night arrival times to understand the current context as "arriving late at night and needing airport pickup"; The system matches the pre-set scenario rule base and determines whether the service trigger threshold has been reached. For example, the system determines that this scenario meets the service scenario of "arriving late at night and needing to be picked up from the airport" and has reached the service trigger threshold; then it proceeds to S30. S30. Make service decisions and push notifications, including: Generate service recommendation schemes and personalized recommendation messages. Following the above, the control agent business center generates a recommendation scheme for "arranging airport pick-up service" and generates personalized recommendation messages.

[0074] Then, the optimal presentation device is determined. In this embodiment, the optimal presentation device is recommended on the core home device—the large TV screen.

[0075] Then, a strong reminder message is pushed to the TV, such as controlling the TV screen to pop up a notification: "Flight XX is delayed and is expected to arrive at 1:00 AM. Transportation is inconvenient late at night. Would you like us to arrange airport pickup service for your family?" Then enter S40; S40, on the TV, service suggestions and operation buttons are prominently displayed in the form of large-screen pop-up windows, and then enter S50; S50: Receive user selection. If the user confirms, proceed to S61; if the user ignores, proceed to S63. S61, The agent calls a third-party service API to complete the transaction and then proceeds to S62; S62. Transaction results are synchronized to the user's mobile phone.

[0076] In this embodiment, the user selects to confirm, specifically, family members can confirm with a single click using the TV remote control, indicating their agreement to book the airport pickup service. Regarding the intelligent agent calling a third-party service API to complete the transaction, this means that after receiving the confirmation instruction, the intelligent agent's business hub immediately calls the third-party airport pickup service API to complete the booking for the airport pickup service.

[0077] Regarding the synchronization of transaction results to the user's mobile phone: In this embodiment, the booking results of the airport pick-up service (such as driver information, estimated arrival time, etc.) will be sent to the mobile phone of the person who made the booking.

[0078] S63. If the user ignores the feedback, record the feedback to optimize the model and exit quietly.

[0079] Through this example, the system can proactively and accurately recommend and complete service transactions on the TV screen before the user is even aware of the need for airport pickup, realizing an intelligent experience of "service finding people" and greatly improving the conversion efficiency and user experience of intelligent services.

[0080] In some embodiments described above, this application proposes an intelligent service recommendation processing method based on multi-device context awareness. This method collects multi-source data, performs deep fusion and contextual understanding on the multi-source data, identifies the real-time context of the target user, and generates a personalized service recommendation scheme based on the context matching and judgment results. However, in practical applications, if the collected multi-source data is too broad or lacks specificity, it may lead to insufficient accuracy in context recognition, thereby affecting the accuracy of service recommendations and user experience.

[0081] In this regard, this application further proposes that the explicit user behavior data includes ticket purchase data and search data; the environmental data includes time data and location data; and the third-party data source includes flight dynamic data and weather data.

[0082] Specifically, the ticket purchase data refers to records generated by users purchasing tickets (such as airline tickets, train tickets, movie tickets, performance tickets, etc.) through various channels. This data can directly reflect users' travel plans, entertainment preferences, or specific needs. It can be achieved by synchronizing data with API interfaces of third-party ticketing platforms (such as airlines, China Railway 12306, cinema chains, etc.) or by parsing ticket confirmation information in authorized emails and SMS messages. Another implementation method is that the system can be integrated into the user's smart device to indirectly obtain this data by monitoring the user's actions in specific ticketing applications. The search data refers to records of user queries made in search engines, app stores, and specific service applications. This data can reveal users' immediate interests, potential needs, or information acquisition intentions. It can be achieved by cooperating with API interfaces of mainstream search engines (such as Baidu, Google, etc.) to obtain the user's authorized search history; or by analyzing search logs within various applications installed on the user's smart device. Another implementation method is that the system can be deployed in a smart home hub to record user query commands through a voice assistant.

[0083] The time data refers to information related to the current time, date, day of the week, and holidays. This data is crucial for determining the time dimension of the user's situation, such as distinguishing between weekdays and weekends, day and night, and specific holidays. This can be achieved by obtaining the local time through the smart device's built-in clock module and calibrating it using the Network Time Protocol (NTP); or by accessing public calendar services to obtain holiday information. Another approach is for the system to determine the user's current time period (e.g., commuting time, meal time, rest time) based on the user's set schedule and the current time. The location data refers to the geographical location information of the user or their smart device, including latitude and longitude, specific location name, and region. This data is crucial for determining the spatial dimension of the user's situation, such as whether the user is at home, at work, at an airport, or in a shopping mall. This can be achieved by obtaining precise latitude and longitude through the smart device's built-in Global Positioning System (GPS) module; or by obtaining a general location through technologies such as Wi-Fi positioning or cellular base station positioning. Another approach is for the system to combine the user's frequently used location markers (e.g., home, office) in map applications to determine the user's location.

[0084] Flight status data refers to real-time information regarding flight takeoffs and landings, delays, cancellations, and gate changes. This data plays a crucial role in identifying user scenarios related to air travel. It can be implemented by connecting to the API interfaces of airlines, airports, or professional flight information service providers (such as FlightAware) to obtain real-time flight status updates. Another approach is for the system to subscribe to airline SMS or email notification services and parse the flight status information within them. Weather data refers to real-time and forecast meteorological information regarding temperature, humidity, precipitation, wind speed, and air quality. This data significantly impacts the assessment of environmental factors in user travel and activity scenarios. It can be implemented by integrating data with meteorological bureaus and commercial weather service providers (such as AccuWeather) via API interfaces to obtain weather forecasts and real-time observation data for specific regions. Another approach is for the system to utilize environmental sensors built into smart home devices to acquire indoor and outdoor temperature and humidity data.

[0085] This application's solution acquires ticket purchase and search data from explicit user behavior data, time and location data from environmental data, and flight status and weather data from third-party data sources. This allows the system to obtain more specific, detailed information closely related to the user's current activities and external environment. When these specific data types are deeply integrated and context-aware, they provide the context-aware module with richer semantic information and more accurate judgment criteria. For example, ticket purchase data directly reveals the user's travel intentions, search data reflects the user's immediate interests, time and location data provide the basic framework of the user's spatiotemporal location, and flight status and weather data supplement key variables of the external environment. These data are interconnected and mutually corroborative, jointly constructing a multi-dimensional, high-precision user context profile. This enables the system to more accurately identify the real-time context of the target user, such as "the user is about to take a flight, but the flight may be delayed and the weather at the destination is bad." This concretization and enrichment of data types significantly improves the accuracy and depth of context awareness, laying a solid foundation for subsequently generating personalized service recommendation solutions.

[0086] The following example illustrates this. Suppose a user books a flight from city A to city B through a ticketing platform; this purchase data is collected by the system. Simultaneously, the system detects that the user has recently been frequently searching for "city B travel guide" and "city B airport transfer service" on search engines; this search data is also recorded. The system obtains the current time data (the night before flight departure) and the user's location data (at home) through a smart device. Furthermore, the system obtains real-time flight status data from a third-party data source, showing an estimated two-hour delay and a 24-hour weather forecast for city B predicting heavy rain. By deeply integrating and understanding the context of this specific purchase data, search data, time data, location data, flight status data, and weather data, the system can accurately identify the target user's real-time scenario as "the user is about to travel, the flight is delayed, the destination has severe weather, and there is a need for transfer." Based on this scenario, the system can generate highly personalized service recommendations, such as "recommending indoor transfer services at city B airport and providing lounge coupons during the delay."

[0087] Through the above technical solution, the system of this invention can acquire user behavior and environmental information from more specific and detailed dimensions, avoiding the ambiguity in context recognition caused by overly generalized data. This clear and detailed understanding of multi-source data types enables the context-aware module to more accurately capture user needs and changes in the external environment, thereby significantly improving the accuracy and depth of real-time context recognition. Ultimately, this allows the system to generate more personalized and targeted service recommendation schemes and recommendations, effectively solving the problem of mismatched recommended services caused by inaccurate context recognition, and greatly improving user satisfaction and acceptance of intelligent service recommendations.

[0088] In other embodiments, this application proposes an intelligent service recommendation processing method based on multi-device context awareness. In the aforementioned embodiments of this application, deep fusion and contextual understanding of multi-source data are proposed to identify the real-time context in which the target user is located. However, in practical applications, when faced with the interweaving of multi-dimensional information such as explicit user behavior data, third-party data sources, and device and environmental data, especially in the context of sudden events and time-sensitive scenarios, how to efficiently and accurately perform deep fusion of these heterogeneous data and extract real-time contexts with clear semantics to avoid ambiguity or lag in context recognition remains a problem that needs to be solved.

[0089] Therefore, this application further proposes a step for deeply fusing and understanding the multi-source data to identify the real-time scenario of the target user, including: when the user's explicit behavior data from the multi-source data is that the user booked a flight ticket through a first terminal, then the system controls the binding of the booked flight information to the first terminal; when the third-party data source of the multi-source data is that the flight information obtained through a second terminal shows an unexpected delay; and when the device and environmental data of the multi-source data is that the system determines the flight will arrive late at night based on time data and location data; then the deep fusion and contextual understanding of the multi-source data involves fusing the flight delay information and the late-night arrival time data to identify the real-time scenario of the target user as arriving late at night and needing to be picked up by the airport.

[0090] In this context, when the explicit user behavior data in the multi-source data refers to user booking a flight ticket through a first terminal, explicit user behavior data refers to user-generated, directly observable, or recordable behavioral data, such as ticket purchases or searches. The first terminal can be a user's smartphone, tablet, or personal computer. The act of booking a flight ticket is completed through the first terminal, and the system can capture this behavior and extract flight information, such as flight number, departure and arrival times, and destination. Binding the flight information of the booked ticket to the first terminal means that the system associates this flight information with the specific device performing the booking operation, so that subsequent contextual awareness and recommendations can be based on the context of that device. This binding can be achieved by storing relevant information on the first terminal or establishing an association between the first terminal and flight information in a cloud service.

[0091] When the third-party data source for the multi-source data is the flight information obtained through a second terminal that shows an unexpected delay, the third-party data source refers to data provided by external services or platforms, such as flight status data, weather data, etc. The second terminal can be a different device from the first terminal, such as a smart speaker, smart TV, or a standalone IoT device, or even an airline or airport server. Obtaining flight information showing an unexpected delay through the second terminal means that the system can monitor abnormal states related to the bound flight in real time from the third-party data source (such as flight tracking services), such as flight delays exceeding a preset threshold. This acquisition can be achieved through API calls, data subscriptions, or web scraping techniques.

[0092] When the system determines that a flight will arrive late at night based on time and location data, the device and environmental data includes the device's own status data (such as battery level and network connection) and environmental data (such as time, location, and weather). Time data can refer to the current system time, the flight's scheduled arrival time, etc. Location data can refer to the user's current location, the airport's location, etc. The system comprehensively analyzes this data, for example, by comparing the flight's estimated arrival time with the current time and combining it with geographical location information, to determine whether the flight will arrive during the late night period (for example, according to a preset rule, 0:00 to 5:00 AM is defined as late night). This determination can be based on preset time interval rules or by using a machine learning model to learn from historical data for prediction.

[0093] The system performs deep fusion and contextual understanding of the multi-source data, fusing flight delay information and late-night arrival time data to identify the target user's real-time scenario as arriving late at night and requiring airport pickup. Deep fusion and contextual understanding involve integrating and analyzing heterogeneous data from different sources, extracting high-level semantic information to accurately identify the user's complex situation. In this specific scenario, the system comprehensively analyzes the user's explicit behavioral data (booked tickets), third-party data sources (flight delay information), and device and environmental data (late-night arrival time). The fusion process can employ rule engines, ontology reasoning, or deep learning-based multimodal data fusion models. Through this fusion, the system understands the complex fact that "the user's booked flight is delayed, and the delay will result in late-night arrival," and thus infers the user's potential real-time scenario of "late-night arrival requiring airport pickup." This scenario recognition is based on the association and reasoning of multiple independent facts, rather than simple data aggregation.

[0094] The proposed solution operates as follows: First, the system collects explicit user behavior data, such as a user's flight booking behavior on a first terminal, and intelligently binds the flight information of the booked flight to that first terminal, thus establishing a preliminary association between the user and a specific flight. Second, the system continuously monitors flight dynamics from third-party data sources. Once it obtains information about a flight that is unexpectedly delayed through a second terminal, it immediately captures this critical event. Simultaneously, the system combines device and environmental data, such as analyzing current time data and the flight's estimated arrival location data, to determine that the flight will arrive late at night due to the delay. After acquiring this multi-source, multi-dimensional data, the proposed solution deeply integrates and contextualizes this discrete but interconnected information. Specifically, the system organically integrates the user's booked flight information, flight delay information provided by third parties, and the late-night arrival time determined through time and location data. This integration is not a simple data overlay, but rather an intelligent analysis that understands the inherent logical relationships and causal chains between these pieces of information. For example, flight delays and late-night arrival together constitute a potential predicament for the user. Through this deep fusion, the system can extract higher-level semantic information from the raw data, thereby accurately identifying the target user's real-time scenario as "arriving late at night and needing airport pickup." This working principle enables the system to accurately capture users' potential needs and problems from complex and dynamically changing data streams, thus laying a solid foundation for subsequent personalized service recommendations. Compared to scenario recognition that relies on a single data source or simple rules, this solution, through deep fusion of multi-source data and contextual understanding, can grasp user scenarios more comprehensively and accurately, especially in dealing with emergencies and time-sensitive scenarios, significantly improving the accuracy and timeliness of scenario recognition.

[0095] The following is a specific example to illustrate this. User Xiao Wang booked a flight from Beijing (A) to Shanghai (B) using his smartphone (the first terminal). The system recorded this explicit user behavior data and linked information such as flight number CA123X, scheduled departure time, and scheduled arrival time to Xiao Wang's smartphone. Before the flight's departure, the system, by connecting to a third-party data source provided by the airline (e.g., through a smart home device or in-vehicle system as a second terminal), detected that flight CA123X was experiencing an unexpected delay due to weather conditions, with an estimated delay of 3 hours. Simultaneously, based on the original scheduled arrival time plus the delay time, combined with current time data and the location data of Shanghai Airport (B), the system determined that flight CA123X would arrive in Shanghai (B) at 2:00 AM, which is considered a late-night arrival. At this point, the system deeply integrated and understood the context of the three pieces of information: "Xiao Wang booked flight CA123X," "Flight CA123X is unexpectedly delayed by 3 hours," and "Flight CA123X arrives late at night." The system identified Xiao Wang's real-time situation as "arriving late at night and needing airport pickup." Based on this scenario, the system of the present invention can further recommend personalized services such as "airport pick-up and drop-off service" and "nearby hotel booking".

[0096] Through the above technical solution, this application effectively addresses the problems of ambiguity or lag in scenario recognition under complex multi-source data environments. Specifically, by deeply integrating and contextualizing explicit user behavior data (such as booked flights), third-party data sources (such as flight delay information), and device and environmental data (such as late-night arrival times), the system can accurately identify specific and time-sensitive real-time scenarios such as "arriving late at night and needing airport pickup" from discrete information. This refined scenario recognition capability enables subsequent service recommendations to more accurately match users' actual needs, avoiding invalid recommendations due to inaccurate scenario judgments, thereby significantly improving the accuracy of intelligent service recommendations and user experience. Especially when dealing with complex scenarios where emergencies (such as flight delays) and environmental factors (such as late at night) overlap, this solution can promptly and accurately capture users' potential pain points and provide them with forward-looking solutions.

[0097] In some of the embodiments described above in this application, a multi-device context-aware intelligent service recommendation processing method is proposed. This method identifies real-time scenarios and makes service recommendations by collecting multi-source data and deeply fusing contextual understanding. However, in its implementation, if recommendations are made solely based on real-time scenarios, it may be difficult to fully capture the personalized differences and deep preferences among multiple users (e.g., family members), resulting in inaccurate recommended services or failure to fully meet the needs of all relevant users.

[0098] In response, this application further proposes that the step of collecting and acquiring multi-source data also includes in-depth analysis of the historical service data of family members on different devices to obtain historical behavioral preference data corresponding to family members; collecting implicit preference data of family members' devices; analyzing the usage habit behavior data of different family members' devices; and storing the historical service data, implicit preference data, and device usage habit behavior data as multidimensional preference data of family members.

[0099] The historical service data of family members on different devices refers to a collection of data related to family members, generated on various smart devices (such as smartphones, tablets, smart speakers, smart TVs, etc.), reflecting their past service usage. This data can include booking records, browsing preferences, payment habits, etc. Booking records specifically refer to purchase details of services such as air tickets, hotels, movie tickets, and food delivery made by users on ticketing platforms, hotel booking websites, and food delivery applications. Browsing preferences can be analyzed through data such as user click behavior, dwell time, and collection records on e-commerce platforms and content aggregation platforms. Payment habits can be obtained from payment platforms, such as users' commonly used payment methods, spending ranges, and preferred merchant types. In-depth analysis of this data can employ various data mining and machine learning techniques. For example, clustering algorithms can be used to group user behavior and identify user groups with similar preferences; association rule mining algorithms can be used to discover purchase correlations between different services; and deep learning models can be used to model time-series data to predict users' future interest trends. Through in-depth analysis, structured and quantified historical behavioral preference data can be obtained, such as user profile tags (e.g., "prefers economy class", "likes Japanese cuisine", "frequently stays in business hotels") or vector representations.

[0100] The implicit preference data of family members' devices refers to user preferences inferred indirectly through device usage patterns, application installations, and social media content, rather than through direct user input or explicit selection. For example, travel application types could refer to ride-hailing apps, navigation apps, hotel booking apps, etc., that the user has installed and frequently uses. Frequently used ride-hailing levels can be inferred based on the vehicle type and service level of the user's historical ride-hailing orders. In addition, by using Natural Language Processing (NLP) technology to analyze the content or chat logs posted by users on social media platforms, keywords related to travel preferences can be extracted, such as "road trip," "island vacation," and "business trip."

[0101] The aforementioned data on the usage habits of different family members refers to the monitoring and analysis of the activity levels and usage periods of various smart devices (such as mobile phones, tablets, and smartwatches) used by family members to understand their device usage patterns and daily routines. The activity level of each user's terminal device can be quantified by indicators such as the number of application launches per day or week, total screen-on time, and data traffic consumption. Usage periods can be determined by statistically analyzing the average usage time or peak activity of the device at different times. For example, data such as the number of application launches and usage time can be obtained through the API interfaces provided by the device's operating system, or by analyzing the device's connection status to a specific network (such as home Wi-Fi) to indirectly infer whether the device is being used at home and for what period.

[0102] Storing historical service data, implicit preference data, and device usage habit data as multidimensional preference data for family members refers to integrating and structuring the various types of preference data mentioned above to form a comprehensive dataset that describes the preferences of family members from multiple perspectives. This can be stored using a relational database, linking different types of data in tables; or using a NoSQL database to store semi-structured or unstructured data; or using a graph database, with family members, devices, preference tags, etc., as nodes and the relationships between them as edges, to facilitate complex relational queries and analyses.

[0103] This application's solution first performs in-depth analysis of family members' historical service data across different devices to obtain explicit behavioral preference data such as booking records, browsing preferences, and payment habits. Simultaneously, the system also collects implicit preference data from family members' devices, such as their frequently used travel app types, ride-hailing rates, and travel preference keywords extracted from social media or chat logs, to capture their unexpressed potential needs. Furthermore, the system analyzes the usage habits and behaviors of different family members' devices, including the activity level and usage time periods of each user's terminal device, to understand their daily routines and device usage patterns. All this processed and analyzed data is ultimately integrated and stored as multidimensional preference data for family members. When the system identifies the target user's real-time context and determines that a service recommendation threshold has been reached, this rich and detailed multidimensional preference data is combined with the real-time context. This combination allows service recommendations to go beyond just the immediate context, deeply considering the long-term preferences, habits, and potential needs of family members, thereby generating more personalized and targeted service recommendation schemes and statements. For example, when the system detects a need for airport pickup, it not only knows that pickup is required, but also recommends high-end private cars for members who prefer comfort and economical carpooling for members who value economy based on multi-dimensional preference data of family members. It also selects the most suitable device for push based on their device usage habits to ensure the effectiveness and acceptance of the recommendations.

[0104] The following is a specific example to illustrate this. Suppose that the system of this invention, through collecting multi-source data, identifies the real-time scenario of a target user (e.g., the father in a family) as "arriving late at night and needing airport pickup." This is because the father booked a flight through a first terminal, the flight information is linked to the first terminal, and the system obtains information about an unexpected flight delay through a second terminal. Simultaneously, the system determines the flight will arrive late at night based on time and location data. Building on this, this application further utilizes multi-dimensional preference data of family members to optimize service recommendations. Specifically, the system has conducted in-depth analysis of the family members' historical service data, finding that the father typically prefers comfortable cars when booking airport pickup services and is accustomed to booking and paying in advance; while the mother's booking records show that she prioritizes cost-effectiveness, often choosing economy or carpooling services. Furthermore, the system also collected data showing that the father has multiple high-end ride-hailing apps installed on his phone, frequently using the "Business Preferred" ride-hailing level, while the mother's social media chat records mention "wanting more economical travel options." Furthermore, by analyzing device usage behavior data, the system discovered that the father's phone is usually in silent mode late at night, but he sets strong reminders for important notifications; the mother's phone is more active late at night, often used for handling family matters. When the system recognizes a scenario of "arriving at the airport late at night and needing to be picked up," it will combine this multi-dimensional preference data to recommend a comfortable ride-hailing service for the father, and considering his habit of keeping his phone silent late at night, push strong reminders through other home devices such as smart speakers or smart TVs; at the same time, it can also provide an economical carpooling option for the mother, and considering her late-night activity, push it directly to her phone.

[0105] Through the aforementioned technical solutions, this application can construct more comprehensive and refined multidimensional preference data for family members, including explicit historical service data, implicit preference data, and device usage habit behavior data. This enables the system to go beyond judgment based on a single scenario after identifying the target user's real-time context, and to gain a deeper understanding of the personalized needs and potential preferences of family members. Therefore, when generating personalized service recommendations and suggestions, it can more accurately match the actual needs and preferences of family members. For example, it can recommend high-end services to members who prefer comfort, and recommend cost-effective services to members who prioritize economy. Furthermore, by analyzing device usage habits, the system can more intelligently decide on the optimal device for service presentation and select appropriate push methods to ensure that recommended information effectively reaches target family members, thereby significantly improving the accuracy of service recommendations, user acceptance, and satisfaction.

[0106] In some of the embodiments described above in this application, a multi-device context awareness approach is proposed to identify real-time scenarios and generate personalized service recommendations by deeply fusing multi-source data and multi-dimensional preference data of family members. However, in practical applications, simply generating a general recommendation and description may not fully meet the diverse needs of family members, nor can it intuitively demonstrate the advantages and disadvantages of different service options, thereby affecting the efficiency and satisfaction of user decision-making.

[0107] In response, this application further proposes that when the current scenario reaches the threshold for triggering service recommendations, a recommendation list containing a predetermined number of differentiated options is generated based on the multidimensional preference data, the current scenario, and the service types available from the third-party service API; wherein each option is accompanied by a brief service description, price range, and degree of matching with the preferences of a certain family member.

[0108] The phrase "when the current scenario reaches the threshold for triggering service recommendations" refers to the system's determination, based on deep fusion of multi-source data and contextual understanding, that the current user's real-time scenario meets preset conditions, necessitating service recommendations. This threshold can be determined based on preset logical rules. For example, it can be triggered when real-time scenario data meets a specific combination of conditions; such as when "flight delay information" and "late-night arrival time data" simultaneously meet preset conditions, airport pickup service recommendations are triggered. Alternatively, the threshold can be dynamically evaluated using a machine learning model. The system trains a classification or regression model that outputs a trigger probability or score based on real-time scenario features (such as time, location, user behavior, device status, etc.). When this probability or score exceeds the preset threshold, service recommendations are triggered.

[0109] "Based on the multidimensional preference data and the current context, combined with the service types available from third-party service APIs" refers to the key information used when generating the recommendation list. The multidimensional preference data refers to comprehensive preference information collected and stored after in-depth analysis of historical service data, implicit preference data, and device usage habit data of family members on different devices during the acquisition of multi-source data, ensuring personalized recommendations. The current context refers to the real-time situation of the target user as identified by the system, ensuring the timeliness and relevance of the recommendations. The service types available from third-party service APIs refer to the specific service types and capabilities that the system can currently provide through interaction with external service providers via interfaces, ensuring the feasibility of the recommendation scheme. For example, the system can maintain a user preference database, storing detailed preference tags for each family member, such as preferences for transportation and dining. Simultaneously, the system can periodically obtain API interface documents from cooperating third-party service providers, parse and store information such as supported service categories, parameter requirements, and price query interfaces, forming a usable service catalog.

[0110] "Generifying a recommendation list containing a predetermined number of differentiated options" means that the system no longer provides a single service recommendation, but instead offers multiple service options with clear distinctions for the user to choose from. Here, "predetermined number" indicates that the number of options in the recommendation list is configurable, for example, it can be set to 3 to 5. These differentiated options can be generated using a rule-based engine. The system filters multiple services that meet the criteria from available third-party services based on the current context and user preferences, and selects a fixed number of services as differentiated options according to preset priority or diversity strategies. Alternatively, they can be generated using a recommendation algorithm. The system uses collaborative filtering, content-based recommendation, or hybrid recommendation algorithms, combined with multi-dimensional user preference data and the current context, to intelligently generate a recommendation list containing multiple different service options from a vast amount of third-party services, ensuring that these options have a certain degree of differentiation in terms of price, service level, brand, etc.

[0111] "Each option includes a brief service description, price range, and match rate with a specific family member's preferences" refers to the detailed information included in each service option in the recommendation list. The brief service description is a concise summary of the service's core features, helping users quickly understand the service content. The price range is the estimated cost range for the service, providing transparent consumption expectations. The match rate with a specific family member's preferences quantifies the degree of fit between the service option and the preferences of that particular family member, which is especially important for joint family decision-making. For example, the brief service description can extract key information from service details obtained from a third-party service API and standardize it. The price range can be obtained by calling the pricing interface of a third-party service API to obtain real-time prices and calculate an estimated price range based on different service parameters. The match rate with a specific family member's preferences can be calculated by maintaining a set of preference tags for each family member and then calculating the overlap or similarity between the features of each service option and the family member's preference tags, expressed as a percentage or rating.

[0112] This application's solution, upon identifying a scenario triggering service recommendations, no longer provides single or general service recommendations. Instead, it fully leverages real-time scenarios identified through multi-source data, deep fusion, and contextual understanding collected in the early stages, as well as multi-dimensional preference data obtained through in-depth analysis of family members' historical service data, implicit preference data, and device usage habits across different devices. Based on this rich and refined information, the system can intelligently generate a recommendation list containing a predetermined number of differentiated service options, combined with the service types available through third-party service APIs. Each option not only includes a brief service description and price range but, more importantly, also provides the degree of matching with the preferences of each family member. This comprehensive recommendation approach makes the recommendations highly personalized, diverse, and transparent, greatly improving the efficiency and satisfaction of users in choosing services in complex scenarios. By combining real-time scenario awareness, multi-dimensional family preference analysis, and third-party service capabilities, this application's solution provides users with a comprehensive and intuitive decision-making support tool, effectively solving the problem that traditional recommendation methods struggle to provide accurate and easy-to-determine recommendations in multi-user, multi-preference scenarios.

[0113] The following example illustrates this. Suppose the system, through deep fusion and contextual understanding of user explicit behavior data, device status data, environmental data, and third-party data sources, identifies the target user's real-time scenario as "arriving late at night and needing airport pickup," and this scenario has reached the threshold for triggering service recommendations. At this point, the system will, based on pre-stored multi-dimensional preference data of family members (e.g., father prefers comfortable cars, mother prefers economical options, and child has no special requirements for car type), and combined with the current scenario (late night, airport location), simultaneously query the service types available from cooperating third-party service APIs (such as ride-hailing platforms and airport transfer service providers). The system will generate a recommendation list containing multiple differentiated options. For example, this list might include: Option 1: "Comfortable private car airport pickup service" provided by a ride-hailing platform, with a brief service description of "spacious vehicle, experienced driver," a price range of "200-250 yuan," and marked "high match with father's preferences (90%), medium match with mother's preferences (60%)." Option 2: Another ride-hailing platform offers an "Economy Express Airport Pickup Service," described as "affordable and fast-responding," priced between 120-150 yuan, and marked as "highly compatible with mother's preferences (85%), moderately compatible with father's preferences (50%)." Option 3: The airport's official "Luxury Business Car Transfer Service," described as "high-end and comfortable, with a dedicated driver," priced between 350-450 yuan, and marked as "moderately compatible with father's preferences (70%), lowly compatible with mother's preferences (30%)." This recommendation list allows family members to intuitively compare the features, costs, and compatibility with their individual preferences of different options, thus helping them make decisions that better meet the collective needs.

[0114] Through the aforementioned technical solution, this application, upon identifying a scenario triggering service recommendations, no longer provides a single or general service recommendation. Instead, it generates a recommendation list containing multiple differentiated options based on multidimensional preference data of family members and real-time scenarios, combined with the actual capabilities of third-party service APIs. Each option is accompanied by a detailed, concise service description, a clear price range, and its match with the preferences of different family members. This allows users, especially families, to intuitively compare the advantages and disadvantages of different service options on a clear and transparent interface, fully considering the personalized needs and budget constraints of family members, thereby making more informed and satisfactory service choices. This solution significantly improves the granularity of service recommendations and the efficiency of user decision-making, effectively addressing the pain point of traditional recommendation methods failing to meet diverse needs in complex family scenarios.

[0115] In some embodiments described above, the system can generate a recommendation list containing a predetermined number of differentiated options based on the current context and multidimensional preference data, determine the optimal service presentation device, push a strong reminder message to that device, and display service suggestions and operation buttons in a prominent manner. However, in multi-user scenarios (e.g., family members), when the recommendation list contains multiple differentiated service options, how to efficiently and intuitively present this complex information to users, effectively guide users in making decisions, and ensure the transparency of the decision-making process and the convenience of transactions are problems that need further solutions.

[0116] In response, this application further proposes that the steps of controlling the generation of personalized service recommendation schemes and recommendations, pushing strong reminder messages to the selected optimal service presentation devices, and displaying service suggestions and operation buttons in a prominent form also include: controlling the terminal's large screen to display multiple service options in a pop-up window, each option being displayed in the form of an independent card, including the service name, key features, estimated price, and selectively labeling the recommended usernames and user preferences; intelligently recommending or highlighting options that match the preferences of the majority or key decision-makers based on user preference matching; if no confirmation selection instruction is received after a predetermined time, the control sends a vote to assist in decision-making according to preset rules or through the terminal; when a confirmation of an option is detected, the control immediately calls the corresponding third-party service API to complete the transaction and synchronizes the transaction result to the mobile terminals of all relevant family members.

[0117] The control terminal displays multiple service options in a large-screen pop-up window. Each option is presented as an independent card, including the service name, key features, estimated price, and optional annotations of recommended usernames and user preferences. This aims to provide a clear and structured information presentation, facilitating quick understanding and comparison of different service options. The large-screen pop-up ensures information prominence, while the card format aids in information segmentation and readability. Annotating usernames and user preferences increases the transparency and persuasiveness of the recommendations, especially in a home setting, helping users understand the reasons for the recommendations. This display method can be triggered by the operating system or specific applications of terminal devices (such as smart TVs, smart speakers with screens, tablets, etc.) receiving a recommendation instruction, resulting in a full-screen or half-screen pop-up interface. The interface adopts a responsive design, automatically adjusting the layout according to the screen size. The data for each service option (name, features, price, preference information) is encapsulated into an independent UI component (card), rendered through a front-end framework or native UI component library. Alternatively, the system generates a dynamic webpage using web technology, which loads as a pop-up in the terminal device's browser or embedded web view. Data for each service option is transmitted to the front end via JSON or other formats. The front-end script dynamically generates a card-style layout and populates the corresponding content. The card design can include various elements such as icons, images, and text to enhance visual appeal.

[0118] Based on user preference matching, options that align with the preferences of the majority or key decision-makers are intelligently recommended or highlighted. This aims to visually guide users to quickly identify options that better suit the needs of the group or key individuals, thereby simplifying the decision-making process and improving efficiency. When rendering service tabs on the front end, specific visual styles can be added to cards that meet the criteria based on the "user preference matching" or "recommendation priority" fields passed from the back end. These styles could include changing the background color, adding borders, displaying a "Recommended" label, or placing the card at the top of the list. Alternatively, when generating the recommendation list, the system calculates a comprehensive recommendation index for each option based on a preset algorithm (e.g., majority vote principle, key decision-maker weight, weighted average of preference scores, etc.). During display, the option with the highest comprehensive recommendation index is highlighted, or a prominent "Intelligent Recommendation" label is displayed on the card.

[0119] If no confirmation selection instruction is received within a predetermined time, the system assists in decision-making by sending a vote through the terminal according to preset rules. This solves the problem of decision-making stagnation caused by differing opinions or delayed responses in multi-user decision-making scenarios. By introducing an automated or semi-automated decision-making assistance mechanism, the smooth progress of the service recommendation process is ensured. An internal timer can be set. When the pop-up window is displayed, the timer starts counting down. If no confirmation selection instruction is received from any user within the preset time (e.g., 5 minutes, 10 minutes), the system will trigger a preset rule (e.g., automatically selecting the option with the highest matching degree, or sending a voting link / notification to the mobile terminals of all relevant family members). Alternatively, the terminal device can start a countdown display while displaying service options. When the countdown ends and no explicit selection is received, the system can automatically send a voting request to the mobile terminals of all associated family members, requesting them to vote on the currently displayed service options and making a decision based on the voting results.

[0120] When a user confirms an option, the system immediately invokes the corresponding third-party service API to complete the transaction and synchronizes the transaction result to the mobile terminals of all relevant family members. This ensures that the service is executed quickly and accurately after the user's decision, and the information synchronization mechanism guarantees that all relevant parties can promptly obtain the transaction status, improving user experience and trust. After receiving the user's "Confirm" button instruction for a service option, the terminal device sends the option's unique identifier and the user's identity information to the backend service. The backend service queries the corresponding third-party service API interface information (such as URL, parameter format, authentication credentials) based on the identifier, constructs an API request, and sends it to the third-party service provider. After the transaction is completed, the third-party service returns the transaction result, which the backend service sends to the mobile terminals of all relevant family members via message queue or push service. Alternatively, after the user confirms the selection, the system will pop up a transaction confirmation interface displaying transaction details. After user confirmation, the system calls the third-party payment API through a secure payment gateway to complete the payment. After successful payment, the system calls the third-party service API (such as a reservation API) to complete the service reservation. The transaction result (such as order number, status, confirmation information) is simultaneously sent to the mobile terminals of all associated family members via instant messaging protocols or SMS / email notifications.

[0121] This application's solution enhances the presentation and decision-making process of service recommendations. When a service recommendation is triggered, the system controls the optimal service presentation device (e.g., a large-screen terminal) to display multiple service options in a clear, structured card format. These cards not only include basic information such as service name, key features, and estimated price, but can also selectively indicate the degree of matching with the preferences of specific family members, ensuring that all family members present can intuitively understand and compare different options. To further assist decision-making, the system intelligently highlights or recommends options that match the preferences of the majority or key decision-makers based on pre-analyzed user preference matching, thereby guiding users to make choices quickly. Considering the potential delays in group decision-making, if the system does not receive a clear selection instruction within a predetermined time, it will automatically make a decision according to preset rules, or assist in decision-making by sending voting requests to the mobile terminals of family members, effectively avoiding stagnation in the decision-making process. Once an option is confirmed, the system immediately calls the corresponding third-party service API to complete the transaction and synchronizes the transaction result to the mobile terminals of all relevant family members in real time, ensuring the convenience, transparency, and consistency of information in the transaction. This integrated approach transforms simple service recommendations into a guided, collaborative decision-making and execution process, significantly improving the service experience in multi-user scenarios.

[0122] The following example illustrates this. Consider a family scenario where the system detects a flight delay and requires airport pickup service. Based on family members' historical service data, implicit preference data, and device usage habits, the system generates a recommendation list containing multiple differentiated pickup service options. At this point, the system controls a pop-up window on the smart TV in the living room (as the large screen) displaying these pickup service options. Each option is presented as an independent card. For example, one card displays "Luxury Car Pickup Service," highlighting key features such as "comfortable and spacious" and "suitable for family travel," with an estimated price of 300 yuan, and labeled "Recommended for: Dad (high preference for comfort)." Another card displays "Economy Car Pickup Service," featuring "high cost-effectiveness" and "fast and convenient," with an estimated price of 150 yuan, and labeled "Recommended for: Mom (prefers affordability)." If the system analysis finds that "Dad" is the key person in the family travel decision-making process, or if the "Luxury Car Pickup Service" has a higher overall family preference match, then that option's card will be highlighted, for example, by changing the border to green or displaying a "Smart Recommendation" label. If a family member has not made a selection on the smart TV within five minutes of the pop-up window appearing, the system will automatically send a voting notification to all family members' mobile devices, such as "Airport delay pick-up service, please select: [Option A] Luxury Car, [Option B] Economy Car". Once a family member confirms the selection of "Luxury Car Pick-up Service" via the smart TV or their mobile device, the system will immediately call the corresponding third-party ride-hailing service API to complete the booking and payment. After the transaction is successful, a confirmation message containing the order number, driver information, estimated arrival time, etc., will be simultaneously pushed to all family members' mobile devices, ensuring that everyone receives the latest status of the pick-up service in a timely manner.

[0123] Through the aforementioned technical solution, this application displays multiple service options in the form of independent cards on the terminal's large screen, clearly labeling the service name, key features, estimated price, and user preferences. This significantly enhances the intuitiveness and readability of information presentation, enabling family members to quickly understand and compare different options. Simultaneously, based on user preference matching, options that align with the preferences of the majority or key decision-makers are intelligently recommended or highlighted, effectively guiding user attention, simplifying the decision-making process, and reducing selection difficulty. Furthermore, if no confirmation instruction is received within a predetermined time, the system can send a vote to assist decision-making according to preset rules or via the terminal, effectively preventing service recommendation process interruptions due to decision delays or differing opinions, thus improving decision-making efficiency and user experience. Once a user confirms an option, the system immediately calls the corresponding third-party service API to complete the transaction and promptly synchronizes the transaction result to the mobile terminals of all relevant family members, ensuring transaction convenience and information transparency, avoiding redundant communication and information asymmetry. Therefore, in multi-user scenarios, this significantly improves the decision-making efficiency, execution efficiency, and user satisfaction of intelligent service recommendations.

[0124] Exemplary device This invention provides an intelligent service recommendation processing device based on multi-device context awareness, aiming to solve the problems of isolated, passive, and lacking context awareness and predictability in existing intelligent device service recommendations. This embodiment combines a multi-source data acquisition module and a context awareness module in a deep fusion manner, and introduces a context matching and matching judgment module, a service decision module, and a push control module, thereby achieving accurate identification of the user's real-time context and proactive service recommendations. This solves the problems of isolated, passive, and lacking context awareness and predictability in intelligent device service recommendations, and achieves the effect of seamless collaboration and proactive service provision between core home devices and personal devices.

[0125] Specifically, such as Figure 3 As shown, the apparatus of this embodiment includes: The multi-source data acquisition module 310 is used to collect and acquire multi-source data, including: user explicit behavior data, device status data, environmental data, and data from third-party data sources; The context awareness module 320 is used to perform deep fusion and context understanding of the multi-source data to identify the real-time context in which the target user is located. The scenario matching and matching judgment module 330 is used to match the real-time scenario of the identified target user with the preset scenario rule library to determine whether the current scenario reaches the threshold for triggering service recommendation; The service decision module 340 is used to generate a personalized service recommendation scheme and recommendation text based on the current scenario when the current scenario reaches the threshold for triggering service recommendation; and to determine the optimal service presentation device. The push control module 350 is used to control the generation of personalized service recommendation schemes and recommendations, push strong reminder messages to the selected optimal service presentation device, and display service suggestions and operation buttons in a prominent form; The operation module 360 ​​is used to control the invocation of the corresponding third-party service API to complete the transaction operation when it detects that the user has confirmed the acceptance of the service operation instruction, and to synchronize the transaction result to the preset user personal device; when it receives the operation instruction that the user chooses to ignore, it controls the system to record and feed back to the preset recommendation model to optimize subsequent recommendations, and then exits, as described above.

[0126] This application's solution comprehensively acquires cross-device data through a multi-source data acquisition module, deeply analyzes user behavior and environmental context through a context-aware module, dynamically evaluates triggering conditions through a context matching and matching judgment module, intelligently generates personalized recommendations and prioritizes the devices for presentation through a service decision module, and ensures strong reminders and smooth interaction through a push control module, thus forming a complete technical closed loop. Compared to traditional solutions that rely solely on the passive response of a single device's historical behavior, this device achieves real-time collaboration and context prediction of multi-device data. For example, in a flight delay scenario, the system can integrate ticketing data, flight status, and location information to proactively push airport pickup service options to the large TV screen, avoiding users switching between different devices and effectively overcoming the shortcomings of service isolation, interaction breaks, and insufficient predictability. Through the above technical solutions, the initiative, accuracy, and user experience continuity of service recommendations are significantly improved.

[0127] Based on the above embodiments, the present invention also provides a server, the principle block diagram of which can be as follows: Figure 4 As shown, the server includes a processor, memory, network interface, display screen, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a multi-device context-aware intelligent service recommendation processing method. The server's database stores the multi-device context-aware intelligent service recommendation processing program.

[0128] Those skilled in the art will understand that Figure 4 The block diagram shown is merely a partial structural diagram related to the present invention and does not constitute a limitation on the server to which the present invention is applied. A specific server may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0129] In one embodiment, a server is provided, comprising a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by one or more processors, the one or more programs including the execution of the methods described above. The server is configured to collect multi-source data, including explicit user behavior data, device status data, environmental data, and data from third-party data sources; perform deep fusion and contextual understanding on the multi-source data to identify the real-time scenario of the target user; match the identified real-time scenario of the target user with a pre-set scenario rule base to determine whether the current scenario reaches a threshold for triggering service recommendations; when the current scenario reaches the threshold for triggering service recommendations, generate a personalized service recommendation scheme and recommendation text based on the current scenario, and determine the optimal service presentation device; control the generation of the personalized service recommendation scheme and recommendation text, push strong reminder messages to the selected optimal service presentation device, and display service suggestions and operation buttons in a prominent manner.

[0130] This embodiment achieves accurate understanding and prediction of users' real-time scenarios by systematically combining key features such as multi-source data acquisition, deep fusion, contextual understanding, scenario matching, and decision-making. Specifically, the server can comprehensively analyze heterogeneous data from devices such as smart TVs and mobile phones. For example, when a user books a flight through a first terminal, the system binds the flight information to that terminal and obtains real-time flight dynamic data from third-party data sources. When an unexpected flight delay is detected and the system determines that the flight will arrive late at night based on time and location data, the system automatically identifies the real-time scenario of needing airport pickup upon arrival late at night. Based on this, the server generates a recommendation list containing differentiated options, such as airport pickup service options matched with multi-dimensional preference data of family members, and decides to use the smart TV as the optimal service presentation device. On the smart TV screen, the system highlights service suggestions and operation buttons in the form of independent cards, such as labeling "Late Night Airport Pickup Service" and highlighting options that match the preferences of key decision-makers, while pushing strong reminder messages to ensure that users are aware of the service in a timely manner.

[0131] This server effectively solves the problems of service isolation, passive recommendations, fragmented interactions, and lack of foresight in the background technologies by achieving deep collaboration and proactive decision-making across device data. Compared to traditional solutions that rely solely on the historical behavior of a single device, this server can predict users' future needs through multi-source data fusion. For example, it can predict late-night transportation inconveniences based on flight delays and proactively provide solutions on core home devices. Simultaneously, by deciding on the optimal service presentation device and completing the transaction loop prominently on that device, it avoids the interaction fragmentation problem of needing to switch to a mobile phone in existing technologies, fully leveraging the potential of large TV screens in service transactions. Overall, this design enables the intelligent system to provide timely and personalized services when users need them most, significantly improving the timeliness, personalization, and user experience of service recommendations.

[0132] In other embodiments, this application proposes a computer-readable storage medium that, when the instructions in the storage medium are executed by the processor of an electronic device, enables the electronic device to perform the above-described method. The core innovation of this embodiment lies in solidifying technical features such as multi-source data acquisition, deep fusion and contextual understanding, scenario matching and decision-making in the form of a computer-readable storage medium. This solves the technical problems of service isolation, passive recommendation, disconnected interaction, and lack of foresight in smart device service recommendations, achieving seamless collaboration and proactive service provision between core home devices and personal devices. Specifically, the instructions stored in the computer-readable storage medium enable the electronic device to acquire multi-source data, including explicit user behavior data, device status data, environmental data, and data from third-party data sources; perform deep fusion and contextual understanding of the multi-source data to identify the real-time scenario of the target user; match the identified real-time scenario with a pre-set scenario rule base to determine whether a threshold for triggering service recommendations has been reached; when the threshold is reached, generate a personalized service recommendation scheme and recommendation text, and decide on the optimal service presentation device; and control the push of strong reminder messages to selected devices to prominently display service suggestions and operation buttons.

[0133] In some of the embodiments described above in this application, a multi-source data acquisition and deep fusion mechanism is proposed. However, during its implementation, it is necessary to ensure that the electronic device can stably execute the complete recommendation process. To address this, this application further proposes an implementation using a computer-readable storage medium. The instructions stored in this medium are configured to be executed by a processor, enabling the electronic device to coordinate the work of the multi-source data acquisition module, the context awareness module, the context matching and matching judgment module, the service decision module, and the push control module. Specifically, the instructions in this storage medium first control the multi-source data acquisition module to acquire user explicit behavior data, device status data, environmental data, and third-party data source information. Second, the context awareness module performs timestamp alignment and semantic association on the data, utilizing machine learning algorithms to achieve contextual understanding. Based on this, the context matching and matching judgment module compares the identified real-time context with the scene rule base to determine whether a recommendation threshold is triggered. When the conditions are met, the service decision module generates a differentiated recommendation list based on the user's multi-dimensional preference data. Finally, the push control module controls the terminal screen to prominently display service suggestions and operation buttons in card format, and calls third-party service APIs to complete transactions according to user instructions.

[0134] This application's solution solidifies the method and process into executable instructions, enabling accurate understanding and prediction of users' real-time scenarios. For example, when the system identifies a "late-night arrival requiring airport pickup" scenario formed by fusing flight delay information with late-night arrival time data, it can generate a recommendation list containing pre-defined differentiated options based on multi-dimensional preference data of family members and determine the optimal service presentation device. This technical solution effectively overcomes the isolation of service recommendations in existing technologies, creating a data loop between the TV screen and mobile devices; it also solves the problem of passive recommendations, allowing the system to proactively trigger service recommendations based on real-time events such as flight delays; furthermore, strong reminder messages and prominent display avoid the interruption of interaction when the transaction process jumps to the mobile phone; ultimately, it achieves predictive response to users' potential needs, such as predicting late-night transportation inconvenience based on delayed flights and providing solutions in advance. One specific implementation example is as follows: When a user books a flight ticket through a first terminal, the system binds the flight information to the device. If a third-party data source shows an unexpected delay and the time data indicates an arrival late at night, an airport pick-up service option matching the preferences of family members will automatically pop up on the smart TV screen. After the user confirms, the system immediately calls the API to complete the transaction and synchronizes the result to the mobile phone, thus providing a seamless service experience when the user needs help the most.

[0135] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for intelligent service recommendation processing based on multi-device context awareness, characterized in that, include: Collect and acquire data from multiple sources, including: explicit user behavior data, device status data, environmental data, and data from third-party data sources; The multi-source data is deeply fused and contextualized to identify the real-time context in which the target user is located; The system identifies the target user's real-time context and matches it against a pre-defined context rule base to determine whether the current context reaches the threshold for triggering service recommendations. When the current scenario reaches the threshold for triggering service recommendations, a personalized service recommendation plan and recommendation text are generated based on the current scenario; and the optimal service presentation device is determined. The system will generate personalized service recommendations and descriptions, push strong reminder messages to selected devices that best present the service, and display service suggestions and operation buttons in a prominent manner.

2. The intelligent service recommendation processing method based on multi-device context awareness according to claim 1, characterized in that, The control system, after generating personalized service recommendation schemes and recommendations, pushing strong reminder messages to selected optimal service presentation devices, and prominently displaying service suggestions and operation buttons, also includes: When the system detects that the user has confirmed acceptance of the service operation instruction, it controls the invocation of the corresponding third-party service API to complete the transaction operation and synchronizes the transaction result to the preset user personal device; When the system receives a user's instruction to ignore an action, it records the instruction, feeds it back to the preset recommendation model to optimize subsequent recommendations, and then exits.

3. The intelligent service recommendation processing method based on multi-device context awareness according to claim 1, characterized in that, The explicit user behavior data includes ticket purchase data and search data; the environmental data includes time data and location data; and the third-party data sources include flight status data and weather data.

4. The intelligent service recommendation processing method based on multi-device context awareness according to claim 1, characterized in that, The steps of performing deep fusion and contextual understanding of the multi-source data to identify the real-time scenario in which the target user is located include: When the explicit user behavior data of the multi-source data is that the user booked a flight ticket through the first terminal, the control binds the flight information of the booked flight ticket to the first terminal; When the third-party data source of the multi-source data is the flight information obtained through a second terminal, and the flight is delayed beyond expectations; And when the device and environmental data from the multi-source data are used by the system to determine that the flight will arrive late at night based on time data and location data; The deep fusion and contextual understanding of the multi-source data are then performed to integrate the flight delay information and late-night arrival time data to identify the real-time scenario of the target user as arriving late at night and needing to be picked up from the airport.

5. The intelligent service recommendation processing method based on multi-device context awareness according to claim 1, characterized in that, The steps for collecting and acquiring multi-source data also include: In-depth analysis of family members’ historical service data across different devices, including booking records, browsing preferences, and payment habits, yields historical behavioral preference data corresponding to family members. The collection of implicit preference data for family members' devices includes the type of travel app, the level of ride-hailing they frequently use, and travel preference keywords mentioned by family members in social media or chat logs; Analyze the usage habits and behaviors of different family members on their devices, including the activity level and usage time of each user's terminal device; The historical service data, implicit preference data, and device usage habit data are stored as multidimensional preference data for family members.

6. The intelligent service recommendation processing method based on multi-device context awareness according to claim 5, characterized in that, When the current scenario reaches the threshold for triggering service recommendations, a personalized service recommendation scheme and recommendation text will be generated based on the current scenario. The steps for determining the optimal service presentation device also include: When the current scenario reaches the threshold for triggering service recommendations, a recommendation list containing a predetermined number of differentiated options is generated based on the multidimensional preference data, the current scenario, and the service types available from the third-party service API. Each option includes a brief service description, a price range, and a degree of matching with the preferences of a particular family member.

7. The intelligent service recommendation processing method based on multi-device context awareness according to claim 6, characterized in that, The steps of controlling the generation of personalized service recommendation schemes and recommendations, pushing strong reminder messages to selected optimal service presentation devices, and displaying service suggestions and operation buttons in a prominent manner also include: The control terminal's large screen pops up multiple service options, each displayed as an independent card, including the service name, key features, estimated price, and can selectively label recommended usernames and user preferences; Based on the degree of matching with user preferences, options that match the preferences of the majority or key decision-makers are intelligently recommended or highlighted; If no confirmation selection instruction is received within the predetermined time, the control will send a vote to assist in decision-making based on preset rules or through the terminal. When a confirmation of an option is detected, the system immediately calls the corresponding third-party service API to complete the transaction and synchronizes the transaction result to the mobile terminals of all relevant family members.

8. A smart service recommendation processing device based on multi-device context awareness, characterized in that, The device includes: The multi-source data acquisition module is used to collect and acquire multi-source data, including: explicit user behavior data, device status data, environmental data, and data from third-party data sources; The context-aware module is used to perform deep fusion and context understanding of the multi-source data to identify the real-time context in which the target user is located. The scenario matching and matching judgment module is used to match the real-time scenario of the identified target user with the pre-set scenario rule library to determine whether the current scenario reaches the threshold for triggering service recommendations. The service decision module is used to generate personalized service recommendation schemes and recommendations based on the current scenario when the current scenario reaches the threshold for triggering service recommendations; and to determine the optimal service presentation device. The push control module is used to control the generation of personalized service recommendation schemes and recommendations, push strong reminder messages to selected optimal service presentation devices, and display service suggestions and operation buttons in a prominent manner.

9. A server, characterized in that, It includes a memory and one or more programs, wherein one or more programs are stored in the memory and configured to be executed by one or more processors, wherein the one or more programs include methods for performing any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, When the instructions in the storage medium are executed by the processor of the electronic device, the electronic device is able to perform the method as described in any one of claims 1-7.

Images